| Abstract: | ABSTRACT Distributions of city sizes are usually characterized by their Pareto index. This, however, turns out to be a rather restrictive view, for the Pareto distribution is known to become singular whenever the Pareto index becomes smaller than one, a case which is fairly frequent in empirical distributions. We show that the introduction of finite Pareto distributions in which city sizes are bounded from above solves a number of difficulties encountered by the rank-size rule and by the unbounded Pareto distribution. Combined with the use of finite Pareto distributions, the green-belt model that has been introduced previously is reexamined. It implies definite constraints for the long-run evolution of urban systems; it is in the cases of countries experiencing a process of fast urbanization that these constraints are of greatest significance. The implications of the model are confronted with empirical evidence concerning the evolution of urban systems in major industrialized countries during the nineteenth and twentieth centuries. |
